Google allows users to disable visible watermarks on AI-generated content

Google is updating its AI policy to allow users to disable visible watermarks, such as the bottom-right sparkle icon, on images, videos, and music generated by its Gemini and Flow tools. While this change aims to give creators more flexibility, Google will continue to embed invisible SynthID watermarks and C2PA metadata to maintain transparency. The toggle will not be available in countries with strict visible watermark regulations, aligning with broader industry shifts toward invisible tracking.

Google AI watermark policy change

• ▪On August 14, 2026, Google announced that it will allow users to disable visible watermarks on AI-generated images, videos, and songs.
• ▪Google is open-sourcing a library called Credentio to enable developers to embed a local validation mechanism for AI-generated content in their applications.
• ▪The setting to turn off visible watermarks will be available in Google Gemini and Google's video editor, Flow, with support for Google Search coming soon.
• ▪When the setting is disabled, Google will remove the sparkle watermark from the bottom-right corner of content generated with its Nano Banana and Omni models.

Invisible watermarking technologies

• ▪Disabling visible watermarks will not affect the invisible SynthID watermarks and C2PA standard-related metadata embedded in Google's AI-generated media.
• ▪Users can still verify if an image or video was AI-generated by using Google Gemini or Google Search to detect the invisible markers.

Industry watermarking practices

• ▪AI image generators from OpenAI and Meta do not apply visible watermarks, with OpenAI relying on SynthID and C2PA, and Meta using its Content Seal standard.
• ▪Anthropic applies invisible watermarks to both AI-generated text and images.

Regulatory compliance considerations

• ▪Anthropic introduced watermarks in text and files generated by its Claude AI model to comply with European Union regulations.
• ▪Google's toggle to disable visible watermarks will not launch in countries that legally require visible watermarks on AI-generated content.

2 sources